Model save
Browse files- README.md +75 -0
- all_results.json +9 -0
- generation_config.json +6 -0
- train_results.json +9 -0
- trainer_state.json +509 -0
README.md
ADDED
|
@@ -0,0 +1,75 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
library_name: transformers
|
| 3 |
+
license: apache-2.0
|
| 4 |
+
base_model: alignment-handbook/zephyr-7b-sft-full
|
| 5 |
+
tags:
|
| 6 |
+
- trl
|
| 7 |
+
- dpo
|
| 8 |
+
- generated_from_trainer
|
| 9 |
+
model-index:
|
| 10 |
+
- name: zephyr-7b-align-scan-4e-07-0.45-cosine-1.0
|
| 11 |
+
results: []
|
| 12 |
+
---
|
| 13 |
+
|
| 14 |
+
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
|
| 15 |
+
should probably proofread and complete it, then remove this comment. -->
|
| 16 |
+
|
| 17 |
+
# zephyr-7b-align-scan-4e-07-0.45-cosine-1.0
|
| 18 |
+
|
| 19 |
+
This model is a fine-tuned version of [alignment-handbook/zephyr-7b-sft-full](https://huggingface.co/alignment-handbook/zephyr-7b-sft-full) on an unknown dataset.
|
| 20 |
+
It achieves the following results on the evaluation set:
|
| 21 |
+
- Loss: 0.6494
|
| 22 |
+
- Rewards/chosen: 0.7541
|
| 23 |
+
- Rewards/rejected: 0.2411
|
| 24 |
+
- Rewards/accuracies: 0.3552
|
| 25 |
+
- Rewards/margins: 0.5130
|
| 26 |
+
- Logps/rejected: -80.5926
|
| 27 |
+
- Logps/chosen: -72.8155
|
| 28 |
+
- Logits/rejected: -2.5245
|
| 29 |
+
- Logits/chosen: -2.5411
|
| 30 |
+
|
| 31 |
+
## Model description
|
| 32 |
+
|
| 33 |
+
More information needed
|
| 34 |
+
|
| 35 |
+
## Intended uses & limitations
|
| 36 |
+
|
| 37 |
+
More information needed
|
| 38 |
+
|
| 39 |
+
## Training and evaluation data
|
| 40 |
+
|
| 41 |
+
More information needed
|
| 42 |
+
|
| 43 |
+
## Training procedure
|
| 44 |
+
|
| 45 |
+
### Training hyperparameters
|
| 46 |
+
|
| 47 |
+
The following hyperparameters were used during training:
|
| 48 |
+
- learning_rate: 4e-07
|
| 49 |
+
- train_batch_size: 8
|
| 50 |
+
- eval_batch_size: 8
|
| 51 |
+
- seed: 42
|
| 52 |
+
- distributed_type: multi-GPU
|
| 53 |
+
- num_devices: 4
|
| 54 |
+
- gradient_accumulation_steps: 2
|
| 55 |
+
- total_train_batch_size: 64
|
| 56 |
+
- total_eval_batch_size: 32
|
| 57 |
+
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
|
| 58 |
+
- lr_scheduler_type: cosine
|
| 59 |
+
- lr_scheduler_warmup_ratio: 0.1
|
| 60 |
+
- num_epochs: 1
|
| 61 |
+
|
| 62 |
+
### Training results
|
| 63 |
+
|
| 64 |
+
| Training Loss | Epoch | Step | Validation Loss | Rewards/chosen | Rewards/rejected | Rewards/accuracies | Rewards/margins | Logps/rejected | Logps/chosen | Logits/rejected | Logits/chosen |
|
| 65 |
+
|:-------------:|:------:|:----:|:---------------:|:--------------:|:----------------:|:------------------:|:---------------:|:--------------:|:------------:|:---------------:|:-------------:|
|
| 66 |
+
| 0.6677 | 0.3484 | 100 | 0.6483 | 0.8134 | 0.3763 | 0.3512 | 0.4372 | -80.2922 | -72.6836 | -2.5394 | -2.5556 |
|
| 67 |
+
| 0.6658 | 0.6969 | 200 | 0.6494 | 0.7541 | 0.2411 | 0.3552 | 0.5130 | -80.5926 | -72.8155 | -2.5245 | -2.5411 |
|
| 68 |
+
|
| 69 |
+
|
| 70 |
+
### Framework versions
|
| 71 |
+
|
| 72 |
+
- Transformers 4.44.2
|
| 73 |
+
- Pytorch 2.4.0
|
| 74 |
+
- Datasets 2.21.0
|
| 75 |
+
- Tokenizers 0.19.1
|
all_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 1.0,
|
| 3 |
+
"total_flos": 0.0,
|
| 4 |
+
"train_loss": 0.6522094729885407,
|
| 5 |
+
"train_runtime": 3372.3295,
|
| 6 |
+
"train_samples": 18340,
|
| 7 |
+
"train_samples_per_second": 5.438,
|
| 8 |
+
"train_steps_per_second": 0.085
|
| 9 |
+
}
|
generation_config.json
ADDED
|
@@ -0,0 +1,6 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_from_model_config": true,
|
| 3 |
+
"bos_token_id": 1,
|
| 4 |
+
"eos_token_id": 2,
|
| 5 |
+
"transformers_version": "4.44.2"
|
| 6 |
+
}
|
train_results.json
ADDED
|
@@ -0,0 +1,9 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"epoch": 1.0,
|
| 3 |
+
"total_flos": 0.0,
|
| 4 |
+
"train_loss": 0.6522094729885407,
|
| 5 |
+
"train_runtime": 3372.3295,
|
| 6 |
+
"train_samples": 18340,
|
| 7 |
+
"train_samples_per_second": 5.438,
|
| 8 |
+
"train_steps_per_second": 0.085
|
| 9 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,509 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 1.0,
|
| 5 |
+
"eval_steps": 100,
|
| 6 |
+
"global_step": 287,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.003484320557491289,
|
| 13 |
+
"grad_norm": 260.1177061514537,
|
| 14 |
+
"learning_rate": 1.3793103448275862e-08,
|
| 15 |
+
"logits/chosen": -2.5345611572265625,
|
| 16 |
+
"logits/rejected": -2.581700563430786,
|
| 17 |
+
"logps/chosen": -60.002105712890625,
|
| 18 |
+
"logps/rejected": -99.98374938964844,
|
| 19 |
+
"loss": 0.6931,
|
| 20 |
+
"rewards/accuracies": 0.0,
|
| 21 |
+
"rewards/chosen": 0.0,
|
| 22 |
+
"rewards/margins": 0.0,
|
| 23 |
+
"rewards/rejected": 0.0,
|
| 24 |
+
"step": 1
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.03484320557491289,
|
| 28 |
+
"grad_norm": 238.95036737182184,
|
| 29 |
+
"learning_rate": 1.3793103448275863e-07,
|
| 30 |
+
"logits/chosen": -2.563187837600708,
|
| 31 |
+
"logits/rejected": -2.5619239807128906,
|
| 32 |
+
"logps/chosen": -59.64890670776367,
|
| 33 |
+
"logps/rejected": -73.38821411132812,
|
| 34 |
+
"loss": 0.6926,
|
| 35 |
+
"rewards/accuracies": 0.2430555522441864,
|
| 36 |
+
"rewards/chosen": 0.004840263165533543,
|
| 37 |
+
"rewards/margins": 0.008125737309455872,
|
| 38 |
+
"rewards/rejected": -0.003285474143922329,
|
| 39 |
+
"step": 10
|
| 40 |
+
},
|
| 41 |
+
{
|
| 42 |
+
"epoch": 0.06968641114982578,
|
| 43 |
+
"grad_norm": 287.8433464848162,
|
| 44 |
+
"learning_rate": 2.7586206896551726e-07,
|
| 45 |
+
"logits/chosen": -2.606386184692383,
|
| 46 |
+
"logits/rejected": -2.56528902053833,
|
| 47 |
+
"logps/chosen": -103.9992446899414,
|
| 48 |
+
"logps/rejected": -94.85326385498047,
|
| 49 |
+
"loss": 0.6826,
|
| 50 |
+
"rewards/accuracies": 0.375,
|
| 51 |
+
"rewards/chosen": 0.05751848220825195,
|
| 52 |
+
"rewards/margins": 0.0391354039311409,
|
| 53 |
+
"rewards/rejected": 0.018383082002401352,
|
| 54 |
+
"step": 20
|
| 55 |
+
},
|
| 56 |
+
{
|
| 57 |
+
"epoch": 0.10452961672473868,
|
| 58 |
+
"grad_norm": 303.36666472427896,
|
| 59 |
+
"learning_rate": 3.999851729426529e-07,
|
| 60 |
+
"logits/chosen": -2.5944175720214844,
|
| 61 |
+
"logits/rejected": -2.5745928287506104,
|
| 62 |
+
"logps/chosen": -82.03970336914062,
|
| 63 |
+
"logps/rejected": -91.46243286132812,
|
| 64 |
+
"loss": 0.6629,
|
| 65 |
+
"rewards/accuracies": 0.3499999940395355,
|
| 66 |
+
"rewards/chosen": 0.20558810234069824,
|
| 67 |
+
"rewards/margins": 0.1706041693687439,
|
| 68 |
+
"rewards/rejected": 0.03498392552137375,
|
| 69 |
+
"step": 30
|
| 70 |
+
},
|
| 71 |
+
{
|
| 72 |
+
"epoch": 0.13937282229965156,
|
| 73 |
+
"grad_norm": 243.63595613939415,
|
| 74 |
+
"learning_rate": 3.982085846069669e-07,
|
| 75 |
+
"logits/chosen": -2.4932851791381836,
|
| 76 |
+
"logits/rejected": -2.4914209842681885,
|
| 77 |
+
"logps/chosen": -77.8083267211914,
|
| 78 |
+
"logps/rejected": -73.65955352783203,
|
| 79 |
+
"loss": 0.645,
|
| 80 |
+
"rewards/accuracies": 0.30000001192092896,
|
| 81 |
+
"rewards/chosen": 0.04954088479280472,
|
| 82 |
+
"rewards/margins": 0.3457469344139099,
|
| 83 |
+
"rewards/rejected": -0.2962060570716858,
|
| 84 |
+
"step": 40
|
| 85 |
+
},
|
| 86 |
+
{
|
| 87 |
+
"epoch": 0.17421602787456447,
|
| 88 |
+
"grad_norm": 209.2280603066799,
|
| 89 |
+
"learning_rate": 3.93496739411485e-07,
|
| 90 |
+
"logits/chosen": -2.519728422164917,
|
| 91 |
+
"logits/rejected": -2.5236926078796387,
|
| 92 |
+
"logps/chosen": -63.118324279785156,
|
| 93 |
+
"logps/rejected": -75.77721405029297,
|
| 94 |
+
"loss": 0.6708,
|
| 95 |
+
"rewards/accuracies": 0.26875001192092896,
|
| 96 |
+
"rewards/chosen": 0.3693724274635315,
|
| 97 |
+
"rewards/margins": 0.22256764769554138,
|
| 98 |
+
"rewards/rejected": 0.14680473506450653,
|
| 99 |
+
"step": 50
|
| 100 |
+
},
|
| 101 |
+
{
|
| 102 |
+
"epoch": 0.20905923344947736,
|
| 103 |
+
"grad_norm": 207.31692334790202,
|
| 104 |
+
"learning_rate": 3.859194147372412e-07,
|
| 105 |
+
"logits/chosen": -2.4840266704559326,
|
| 106 |
+
"logits/rejected": -2.477837324142456,
|
| 107 |
+
"logps/chosen": -71.02902221679688,
|
| 108 |
+
"logps/rejected": -66.88030242919922,
|
| 109 |
+
"loss": 0.6573,
|
| 110 |
+
"rewards/accuracies": 0.3375000059604645,
|
| 111 |
+
"rewards/chosen": 0.9798685908317566,
|
| 112 |
+
"rewards/margins": 0.2921520173549652,
|
| 113 |
+
"rewards/rejected": 0.687716543674469,
|
| 114 |
+
"step": 60
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.24390243902439024,
|
| 118 |
+
"grad_norm": 237.196777331744,
|
| 119 |
+
"learning_rate": 3.7558882264746717e-07,
|
| 120 |
+
"logits/chosen": -2.5007472038269043,
|
| 121 |
+
"logits/rejected": -2.4955484867095947,
|
| 122 |
+
"logps/chosen": -61.67304611206055,
|
| 123 |
+
"logps/rejected": -66.74317169189453,
|
| 124 |
+
"loss": 0.6486,
|
| 125 |
+
"rewards/accuracies": 0.32499998807907104,
|
| 126 |
+
"rewards/chosen": 1.153420090675354,
|
| 127 |
+
"rewards/margins": 0.39587074518203735,
|
| 128 |
+
"rewards/rejected": 0.7575494050979614,
|
| 129 |
+
"step": 70
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.2787456445993031,
|
| 133 |
+
"grad_norm": 258.03955447258926,
|
| 134 |
+
"learning_rate": 3.6265794814711307e-07,
|
| 135 |
+
"logits/chosen": -2.449399709701538,
|
| 136 |
+
"logits/rejected": -2.439444065093994,
|
| 137 |
+
"logps/chosen": -73.67487335205078,
|
| 138 |
+
"logps/rejected": -76.30928039550781,
|
| 139 |
+
"loss": 0.671,
|
| 140 |
+
"rewards/accuracies": 0.3125,
|
| 141 |
+
"rewards/chosen": 0.8954970240592957,
|
| 142 |
+
"rewards/margins": 0.3529408872127533,
|
| 143 |
+
"rewards/rejected": 0.5425562262535095,
|
| 144 |
+
"step": 80
|
| 145 |
+
},
|
| 146 |
+
{
|
| 147 |
+
"epoch": 0.313588850174216,
|
| 148 |
+
"grad_norm": 242.91718492869416,
|
| 149 |
+
"learning_rate": 3.4731828363876876e-07,
|
| 150 |
+
"logits/chosen": -2.4762511253356934,
|
| 151 |
+
"logits/rejected": -2.4902822971343994,
|
| 152 |
+
"logps/chosen": -64.53350830078125,
|
| 153 |
+
"logps/rejected": -69.27278900146484,
|
| 154 |
+
"loss": 0.6857,
|
| 155 |
+
"rewards/accuracies": 0.3062500059604645,
|
| 156 |
+
"rewards/chosen": 0.6519443988800049,
|
| 157 |
+
"rewards/margins": 0.27337905764579773,
|
| 158 |
+
"rewards/rejected": 0.37856537103652954,
|
| 159 |
+
"step": 90
|
| 160 |
+
},
|
| 161 |
+
{
|
| 162 |
+
"epoch": 0.34843205574912894,
|
| 163 |
+
"grad_norm": 261.21801117417044,
|
| 164 |
+
"learning_rate": 3.297969931252667e-07,
|
| 165 |
+
"logits/chosen": -2.466578483581543,
|
| 166 |
+
"logits/rejected": -2.466921329498291,
|
| 167 |
+
"logps/chosen": -73.65306091308594,
|
| 168 |
+
"logps/rejected": -80.45304870605469,
|
| 169 |
+
"loss": 0.6677,
|
| 170 |
+
"rewards/accuracies": 0.3375000059604645,
|
| 171 |
+
"rewards/chosen": 0.7281513214111328,
|
| 172 |
+
"rewards/margins": 0.4587801396846771,
|
| 173 |
+
"rewards/rejected": 0.2693712115287781,
|
| 174 |
+
"step": 100
|
| 175 |
+
},
|
| 176 |
+
{
|
| 177 |
+
"epoch": 0.34843205574912894,
|
| 178 |
+
"eval_logits/chosen": -2.5556371212005615,
|
| 179 |
+
"eval_logits/rejected": -2.539428472518921,
|
| 180 |
+
"eval_logps/chosen": -72.68360900878906,
|
| 181 |
+
"eval_logps/rejected": -80.29220581054688,
|
| 182 |
+
"eval_loss": 0.6482642889022827,
|
| 183 |
+
"eval_rewards/accuracies": 0.3511904776096344,
|
| 184 |
+
"eval_rewards/chosen": 0.8134303092956543,
|
| 185 |
+
"eval_rewards/margins": 0.43716490268707275,
|
| 186 |
+
"eval_rewards/rejected": 0.376265287399292,
|
| 187 |
+
"eval_runtime": 116.9271,
|
| 188 |
+
"eval_samples_per_second": 17.105,
|
| 189 |
+
"eval_steps_per_second": 0.539,
|
| 190 |
+
"step": 100
|
| 191 |
+
},
|
| 192 |
+
{
|
| 193 |
+
"epoch": 0.3832752613240418,
|
| 194 |
+
"grad_norm": 287.07453268461984,
|
| 195 |
+
"learning_rate": 3.103535481540892e-07,
|
| 196 |
+
"logits/chosen": -2.494990587234497,
|
| 197 |
+
"logits/rejected": -2.457904100418091,
|
| 198 |
+
"logps/chosen": -71.97386169433594,
|
| 199 |
+
"logps/rejected": -62.46455764770508,
|
| 200 |
+
"loss": 0.6685,
|
| 201 |
+
"rewards/accuracies": 0.2750000059604645,
|
| 202 |
+
"rewards/chosen": 0.5382742881774902,
|
| 203 |
+
"rewards/margins": 0.29256415367126465,
|
| 204 |
+
"rewards/rejected": 0.24571006000041962,
|
| 205 |
+
"step": 110
|
| 206 |
+
},
|
| 207 |
+
{
|
| 208 |
+
"epoch": 0.4181184668989547,
|
| 209 |
+
"grad_norm": 197.14505664650054,
|
| 210 |
+
"learning_rate": 2.8927588532163986e-07,
|
| 211 |
+
"logits/chosen": -2.5188653469085693,
|
| 212 |
+
"logits/rejected": -2.486799955368042,
|
| 213 |
+
"logps/chosen": -76.14109802246094,
|
| 214 |
+
"logps/rejected": -66.33953857421875,
|
| 215 |
+
"loss": 0.6406,
|
| 216 |
+
"rewards/accuracies": 0.3125,
|
| 217 |
+
"rewards/chosen": 0.5942131876945496,
|
| 218 |
+
"rewards/margins": 0.44479331374168396,
|
| 219 |
+
"rewards/rejected": 0.14941997826099396,
|
| 220 |
+
"step": 120
|
| 221 |
+
},
|
| 222 |
+
{
|
| 223 |
+
"epoch": 0.4529616724738676,
|
| 224 |
+
"grad_norm": 309.7749734748652,
|
| 225 |
+
"learning_rate": 2.6687614224062655e-07,
|
| 226 |
+
"logits/chosen": -2.5562925338745117,
|
| 227 |
+
"logits/rejected": -2.537079095840454,
|
| 228 |
+
"logps/chosen": -82.76345825195312,
|
| 229 |
+
"logps/rejected": -87.87675476074219,
|
| 230 |
+
"loss": 0.661,
|
| 231 |
+
"rewards/accuracies": 0.3812499940395355,
|
| 232 |
+
"rewards/chosen": 0.5027960538864136,
|
| 233 |
+
"rewards/margins": 0.6685007810592651,
|
| 234 |
+
"rewards/rejected": -0.16570481657981873,
|
| 235 |
+
"step": 130
|
| 236 |
+
},
|
| 237 |
+
{
|
| 238 |
+
"epoch": 0.4878048780487805,
|
| 239 |
+
"grad_norm": 207.92836576726873,
|
| 240 |
+
"learning_rate": 2.434860351163114e-07,
|
| 241 |
+
"logits/chosen": -2.4571573734283447,
|
| 242 |
+
"logits/rejected": -2.445655345916748,
|
| 243 |
+
"logps/chosen": -79.63760375976562,
|
| 244 |
+
"logps/rejected": -70.80236053466797,
|
| 245 |
+
"loss": 0.6301,
|
| 246 |
+
"rewards/accuracies": 0.3687500059604645,
|
| 247 |
+
"rewards/chosen": 0.678521454334259,
|
| 248 |
+
"rewards/margins": 0.6956696510314941,
|
| 249 |
+
"rewards/rejected": -0.017148202285170555,
|
| 250 |
+
"step": 140
|
| 251 |
+
},
|
| 252 |
+
{
|
| 253 |
+
"epoch": 0.5226480836236934,
|
| 254 |
+
"grad_norm": 266.02287575831133,
|
| 255 |
+
"learning_rate": 2.194519463847738e-07,
|
| 256 |
+
"logits/chosen": -2.522045612335205,
|
| 257 |
+
"logits/rejected": -2.4781124591827393,
|
| 258 |
+
"logps/chosen": -78.24339294433594,
|
| 259 |
+
"logps/rejected": -79.53880310058594,
|
| 260 |
+
"loss": 0.6665,
|
| 261 |
+
"rewards/accuracies": 0.30000001192092896,
|
| 262 |
+
"rewards/chosen": 0.5584267377853394,
|
| 263 |
+
"rewards/margins": 0.5530349016189575,
|
| 264 |
+
"rewards/rejected": 0.005391845945268869,
|
| 265 |
+
"step": 150
|
| 266 |
+
},
|
| 267 |
+
{
|
| 268 |
+
"epoch": 0.5574912891986062,
|
| 269 |
+
"grad_norm": 250.7616411510584,
|
| 270 |
+
"learning_rate": 1.951297951600021e-07,
|
| 271 |
+
"logits/chosen": -2.4823291301727295,
|
| 272 |
+
"logits/rejected": -2.5029988288879395,
|
| 273 |
+
"logps/chosen": -63.5211067199707,
|
| 274 |
+
"logps/rejected": -71.80608367919922,
|
| 275 |
+
"loss": 0.6696,
|
| 276 |
+
"rewards/accuracies": 0.2750000059604645,
|
| 277 |
+
"rewards/chosen": 0.45168429613113403,
|
| 278 |
+
"rewards/margins": 0.3766103982925415,
|
| 279 |
+
"rewards/rejected": 0.07507390528917313,
|
| 280 |
+
"step": 160
|
| 281 |
+
},
|
| 282 |
+
{
|
| 283 |
+
"epoch": 0.5923344947735192,
|
| 284 |
+
"grad_norm": 247.2037562642855,
|
| 285 |
+
"learning_rate": 1.7087976645299907e-07,
|
| 286 |
+
"logits/chosen": -2.485832691192627,
|
| 287 |
+
"logits/rejected": -2.471374034881592,
|
| 288 |
+
"logps/chosen": -68.3868637084961,
|
| 289 |
+
"logps/rejected": -76.93021392822266,
|
| 290 |
+
"loss": 0.6169,
|
| 291 |
+
"rewards/accuracies": 0.30000001192092896,
|
| 292 |
+
"rewards/chosen": 0.2612365484237671,
|
| 293 |
+
"rewards/margins": 0.5385117530822754,
|
| 294 |
+
"rewards/rejected": -0.2772751450538635,
|
| 295 |
+
"step": 170
|
| 296 |
+
},
|
| 297 |
+
{
|
| 298 |
+
"epoch": 0.627177700348432,
|
| 299 |
+
"grad_norm": 253.43980546820382,
|
| 300 |
+
"learning_rate": 1.4706097721752126e-07,
|
| 301 |
+
"logits/chosen": -2.5348095893859863,
|
| 302 |
+
"logits/rejected": -2.523033618927002,
|
| 303 |
+
"logps/chosen": -91.01544189453125,
|
| 304 |
+
"logps/rejected": -86.55479431152344,
|
| 305 |
+
"loss": 0.6875,
|
| 306 |
+
"rewards/accuracies": 0.33125001192092896,
|
| 307 |
+
"rewards/chosen": 0.37877795100212097,
|
| 308 |
+
"rewards/margins": 0.398730605840683,
|
| 309 |
+
"rewards/rejected": -0.019952651113271713,
|
| 310 |
+
"step": 180
|
| 311 |
+
},
|
| 312 |
+
{
|
| 313 |
+
"epoch": 0.662020905923345,
|
| 314 |
+
"grad_norm": 199.96875567633583,
|
| 315 |
+
"learning_rate": 1.240261582126029e-07,
|
| 316 |
+
"logits/chosen": -2.5323076248168945,
|
| 317 |
+
"logits/rejected": -2.5228934288024902,
|
| 318 |
+
"logps/chosen": -70.24234771728516,
|
| 319 |
+
"logps/rejected": -81.30900573730469,
|
| 320 |
+
"loss": 0.641,
|
| 321 |
+
"rewards/accuracies": 0.3187499940395355,
|
| 322 |
+
"rewards/chosen": 0.36411014199256897,
|
| 323 |
+
"rewards/margins": 0.26548105478286743,
|
| 324 |
+
"rewards/rejected": 0.09862907975912094,
|
| 325 |
+
"step": 190
|
| 326 |
+
},
|
| 327 |
+
{
|
| 328 |
+
"epoch": 0.6968641114982579,
|
| 329 |
+
"grad_norm": 331.81322846783684,
|
| 330 |
+
"learning_rate": 1.0211643043778293e-07,
|
| 331 |
+
"logits/chosen": -2.5657148361206055,
|
| 332 |
+
"logits/rejected": -2.5730998516082764,
|
| 333 |
+
"logps/chosen": -88.71566009521484,
|
| 334 |
+
"logps/rejected": -91.1668930053711,
|
| 335 |
+
"loss": 0.6658,
|
| 336 |
+
"rewards/accuracies": 0.375,
|
| 337 |
+
"rewards/chosen": 0.8733166456222534,
|
| 338 |
+
"rewards/margins": 0.4756090044975281,
|
| 339 |
+
"rewards/rejected": 0.3977075219154358,
|
| 340 |
+
"step": 200
|
| 341 |
+
},
|
| 342 |
+
{
|
| 343 |
+
"epoch": 0.6968641114982579,
|
| 344 |
+
"eval_logits/chosen": -2.541111707687378,
|
| 345 |
+
"eval_logits/rejected": -2.5245211124420166,
|
| 346 |
+
"eval_logps/chosen": -72.81554412841797,
|
| 347 |
+
"eval_logps/rejected": -80.59257507324219,
|
| 348 |
+
"eval_loss": 0.6494360566139221,
|
| 349 |
+
"eval_rewards/accuracies": 0.3551587164402008,
|
| 350 |
+
"eval_rewards/chosen": 0.7540590763092041,
|
| 351 |
+
"eval_rewards/margins": 0.5129595994949341,
|
| 352 |
+
"eval_rewards/rejected": 0.24109944701194763,
|
| 353 |
+
"eval_runtime": 113.9336,
|
| 354 |
+
"eval_samples_per_second": 17.554,
|
| 355 |
+
"eval_steps_per_second": 0.553,
|
| 356 |
+
"step": 200
|
| 357 |
+
},
|
| 358 |
+
{
|
| 359 |
+
"epoch": 0.7317073170731707,
|
| 360 |
+
"grad_norm": 387.8686320796793,
|
| 361 |
+
"learning_rate": 8.165625349643729e-08,
|
| 362 |
+
"logits/chosen": -2.5471763610839844,
|
| 363 |
+
"logits/rejected": -2.5208678245544434,
|
| 364 |
+
"logps/chosen": -67.77056884765625,
|
| 365 |
+
"logps/rejected": -63.1707649230957,
|
| 366 |
+
"loss": 0.6465,
|
| 367 |
+
"rewards/accuracies": 0.36250001192092896,
|
| 368 |
+
"rewards/chosen": 0.6968271136283875,
|
| 369 |
+
"rewards/margins": 0.5916789770126343,
|
| 370 |
+
"rewards/rejected": 0.10514805465936661,
|
| 371 |
+
"step": 210
|
| 372 |
+
},
|
| 373 |
+
{
|
| 374 |
+
"epoch": 0.7665505226480837,
|
| 375 |
+
"grad_norm": 195.53605803419836,
|
| 376 |
+
"learning_rate": 6.294862069654417e-08,
|
| 377 |
+
"logits/chosen": -2.573523759841919,
|
| 378 |
+
"logits/rejected": -2.5524840354919434,
|
| 379 |
+
"logps/chosen": -71.91072082519531,
|
| 380 |
+
"logps/rejected": -70.82460021972656,
|
| 381 |
+
"loss": 0.6589,
|
| 382 |
+
"rewards/accuracies": 0.28125,
|
| 383 |
+
"rewards/chosen": 0.6963558197021484,
|
| 384 |
+
"rewards/margins": 0.3189951777458191,
|
| 385 |
+
"rewards/rejected": 0.37736067175865173,
|
| 386 |
+
"step": 220
|
| 387 |
+
},
|
| 388 |
+
{
|
| 389 |
+
"epoch": 0.8013937282229965,
|
| 390 |
+
"grad_norm": 279.18152746545553,
|
| 391 |
+
"learning_rate": 4.6270572044293563e-08,
|
| 392 |
+
"logits/chosen": -2.5709030628204346,
|
| 393 |
+
"logits/rejected": -2.5481770038604736,
|
| 394 |
+
"logps/chosen": -88.14967346191406,
|
| 395 |
+
"logps/rejected": -88.01075744628906,
|
| 396 |
+
"loss": 0.6489,
|
| 397 |
+
"rewards/accuracies": 0.3687500059604645,
|
| 398 |
+
"rewards/chosen": 0.8329347372055054,
|
| 399 |
+
"rewards/margins": 0.7707937359809875,
|
| 400 |
+
"rewards/rejected": 0.06214110180735588,
|
| 401 |
+
"step": 230
|
| 402 |
+
},
|
| 403 |
+
{
|
| 404 |
+
"epoch": 0.8362369337979094,
|
| 405 |
+
"grad_norm": 218.8441311845723,
|
| 406 |
+
"learning_rate": 3.186909157830124e-08,
|
| 407 |
+
"logits/chosen": -2.563115119934082,
|
| 408 |
+
"logits/rejected": -2.5264599323272705,
|
| 409 |
+
"logps/chosen": -85.11534881591797,
|
| 410 |
+
"logps/rejected": -79.63955688476562,
|
| 411 |
+
"loss": 0.6431,
|
| 412 |
+
"rewards/accuracies": 0.3499999940395355,
|
| 413 |
+
"rewards/chosen": 0.7318143248558044,
|
| 414 |
+
"rewards/margins": 0.5083704590797424,
|
| 415 |
+
"rewards/rejected": 0.223443865776062,
|
| 416 |
+
"step": 240
|
| 417 |
+
},
|
| 418 |
+
{
|
| 419 |
+
"epoch": 0.8710801393728222,
|
| 420 |
+
"grad_norm": 242.6767041893939,
|
| 421 |
+
"learning_rate": 1.9957449800512527e-08,
|
| 422 |
+
"logits/chosen": -2.5801711082458496,
|
| 423 |
+
"logits/rejected": -2.544325113296509,
|
| 424 |
+
"logps/chosen": -93.85467529296875,
|
| 425 |
+
"logps/rejected": -89.73026275634766,
|
| 426 |
+
"loss": 0.5901,
|
| 427 |
+
"rewards/accuracies": 0.40625,
|
| 428 |
+
"rewards/chosen": 0.7175648212432861,
|
| 429 |
+
"rewards/margins": 0.6203989386558533,
|
| 430 |
+
"rewards/rejected": 0.09716588258743286,
|
| 431 |
+
"step": 250
|
| 432 |
+
},
|
| 433 |
+
{
|
| 434 |
+
"epoch": 0.9059233449477352,
|
| 435 |
+
"grad_norm": 197.6846790605052,
|
| 436 |
+
"learning_rate": 1.0712045368478117e-08,
|
| 437 |
+
"logits/chosen": -2.4915573596954346,
|
| 438 |
+
"logits/rejected": -2.50749135017395,
|
| 439 |
+
"logps/chosen": -57.76538848876953,
|
| 440 |
+
"logps/rejected": -65.16616821289062,
|
| 441 |
+
"loss": 0.6473,
|
| 442 |
+
"rewards/accuracies": 0.2874999940395355,
|
| 443 |
+
"rewards/chosen": 0.5777884721755981,
|
| 444 |
+
"rewards/margins": 0.3698353171348572,
|
| 445 |
+
"rewards/rejected": 0.20795314013957977,
|
| 446 |
+
"step": 260
|
| 447 |
+
},
|
| 448 |
+
{
|
| 449 |
+
"epoch": 0.9407665505226481,
|
| 450 |
+
"grad_norm": 269.9944966251555,
|
| 451 |
+
"learning_rate": 4.269792820155782e-09,
|
| 452 |
+
"logits/chosen": -2.588334560394287,
|
| 453 |
+
"logits/rejected": -2.5887222290039062,
|
| 454 |
+
"logps/chosen": -67.41829681396484,
|
| 455 |
+
"logps/rejected": -82.52043151855469,
|
| 456 |
+
"loss": 0.6161,
|
| 457 |
+
"rewards/accuracies": 0.3499999940395355,
|
| 458 |
+
"rewards/chosen": 0.6578438878059387,
|
| 459 |
+
"rewards/margins": 0.6527446508407593,
|
| 460 |
+
"rewards/rejected": 0.005099198315292597,
|
| 461 |
+
"step": 270
|
| 462 |
+
},
|
| 463 |
+
{
|
| 464 |
+
"epoch": 0.975609756097561,
|
| 465 |
+
"grad_norm": 209.58164118406862,
|
| 466 |
+
"learning_rate": 7.260950162363721e-10,
|
| 467 |
+
"logits/chosen": -2.499262809753418,
|
| 468 |
+
"logits/rejected": -2.475175380706787,
|
| 469 |
+
"logps/chosen": -66.43128967285156,
|
| 470 |
+
"logps/rejected": -70.41940307617188,
|
| 471 |
+
"loss": 0.6212,
|
| 472 |
+
"rewards/accuracies": 0.3062500059604645,
|
| 473 |
+
"rewards/chosen": 0.6313899159431458,
|
| 474 |
+
"rewards/margins": 0.44463640451431274,
|
| 475 |
+
"rewards/rejected": 0.18675348162651062,
|
| 476 |
+
"step": 280
|
| 477 |
+
},
|
| 478 |
+
{
|
| 479 |
+
"epoch": 1.0,
|
| 480 |
+
"step": 287,
|
| 481 |
+
"total_flos": 0.0,
|
| 482 |
+
"train_loss": 0.6522094729885407,
|
| 483 |
+
"train_runtime": 3372.3295,
|
| 484 |
+
"train_samples_per_second": 5.438,
|
| 485 |
+
"train_steps_per_second": 0.085
|
| 486 |
+
}
|
| 487 |
+
],
|
| 488 |
+
"logging_steps": 10,
|
| 489 |
+
"max_steps": 287,
|
| 490 |
+
"num_input_tokens_seen": 0,
|
| 491 |
+
"num_train_epochs": 1,
|
| 492 |
+
"save_steps": 100,
|
| 493 |
+
"stateful_callbacks": {
|
| 494 |
+
"TrainerControl": {
|
| 495 |
+
"args": {
|
| 496 |
+
"should_epoch_stop": false,
|
| 497 |
+
"should_evaluate": false,
|
| 498 |
+
"should_log": false,
|
| 499 |
+
"should_save": true,
|
| 500 |
+
"should_training_stop": true
|
| 501 |
+
},
|
| 502 |
+
"attributes": {}
|
| 503 |
+
}
|
| 504 |
+
},
|
| 505 |
+
"total_flos": 0.0,
|
| 506 |
+
"train_batch_size": 8,
|
| 507 |
+
"trial_name": null,
|
| 508 |
+
"trial_params": null
|
| 509 |
+
}
|